Japan’s SoftBank provides AI strategy update

• SoftBank has reported much improved financials for the first three quarters of its current financial year
• The Japanese telco used its latest earnings presentation to update its wide-ranging AI strategy
• It aims to provide Japan with the “next generation of social infrastructure”, supported by distributed datacentres that include AI-RAN deployments 

Japanese communications service provider SoftBank Corp. provided a revealing update on its wide-ranging AI strategy as it reported much improved sales and profits for the first nine months of its current financial year. 

The telco reported a 7% year-on-year increase in revenues to 4,811.5bn yen ($31.6bn) for the nine months that ended 31 December 2024 (the first three quarters of its fiscal year), and a 12% increase in operating profits to 821.9bn yen ($5.4bn) for the period, putting it slightly ahead of its sales plan and a long way ahead of its full fiscal year profit expectations. 

The increase was helped by a return to growth in the company’s consumer mobile services unit, which recorded a 2% increase in revenues to 1,177bn yen ($7.72bn), following three consecutive years of declining sales.

SoftBank ended December 2024 with 31.3 million mobile customers across its various brands. 

SoftBank’s enterprise services revenues (mobile, fixed line, business solutions) also trended up, increasing by 10% year on year to 673.6bn yen ($4.42bn) for the first three quarters of the fiscal year. 

The operator, like its peers in Japan and the Asia-Pacific region, is developing a revamped operational and business strategy that hinges on AI, driven by the increasing involvement of its parent company, SoftBank Group, in global AI initiatives such as the $500bn Stargate Project in the US and the growing relationship between SoftBank Group and ChatGPT developer OpenAI (which, speculation has it, could include a sizeable investment). 

SoftBank Corp. isn’t just watching its parent company do the heavy lifting, though – the telco is heavily involved in the ongoing developments with OpenAI, including the development, promotion and deployment of an “advanced enterprise AI” system called Cristal Intelligence, which is being designed to “securely integrate the systems and data of individual enterprises in a way that is customised specifically for each company”. SoftBank Group has pledged to spend US$3bn annually to deploy OpenAI’s solutions across its group companies (including SoftBank Corp.), “making it the first company in the world to integrate Cristal Intelligence at scale, as well as deploying existing tools like ChatGPT Enterprise to employees across the entire group.” 

In addition, SoftBank and OpenAI are establishing a joint venture (SB OpenAI Japan) to “accelerate the deployment of Cristal Intelligence customised for Japan-based companies”, and SoftBank Corp. has a stake in that venture.      

For its part, SoftBank Corp. is working towards the deployment of an AI-RAN architecture following the development of an AI-RAN solution, dubbed AITRAS, with Nvidia late last year. SoftBank used its earnings presentation to promote the virtues of AI-RAN, which (according to SoftBank) include greater energy efficiency, the ability to develop new revenue-generating services, the instantiation of distributed compute resources for AI workload processing and enhanced mobile services. 

The telco is also investing in and building a major AI datacentre facility on the site of a former plant run by Japanese tech firm Sharp in Sakai, located roughly in the centre of Japan. SoftBank acquired land and buildings on the site for 100bn yen ($657m) and expects to have the datacentre up and running in 2026. It’s worth noting that it will have rival Japanese telco KDDI as a neighbour at that location, as KDDI also acquired land from Sharp at its Sakai location in order to build what it referred to as “the largest AI datacentre in Asia”.

SoftBank refers to that facility as its ‘Brain datacentre’, one that has a large capacity, is able to house large banks of high-performance compute systems (and even quantum computers) and can support the development of third-party AI systems and applications as well as help with SoftBank’s own large language model (LLM) training and sovereign cloud service developments. 

That Brain datacentre will support SoftBank and its partners and customers directly, but it will also support a range of other distributed SoftBank datacentres, namely ‘Regional brain’, ‘Sub-regional brain’ and edge computing facilities, with the latter eventually comprising (at least in part) the AI-RAN deployments. 

Ultimately, SoftBank aims to “master the use of AI” and provide Japan with the “next generation of social infrastructure” underpinned by its datacentre facilities, digital platforms and communications networks.

- Ray Le Maistre, Editorial Director, TelecomTV